AI (decision to buy this sim)

rapid · February 25, 2009

LB2 will always have a place on my shelf BS is now king of the hill, But the A! in LB2 for its day wasn't all that bad.

The ground AI would shoot at you when you landed to drop troupes off and pick troupes up you could see your soldiers run off and on your hewy.

BS is Great and will only improve over time

The king is dead...long live the King...........:smilewink:

FADM Stern GNSF · February 25, 2009

Well, I can understand the piont of wanting super smart AI and dynamic missions, but there is always a trade-off. DCS has put a lot into a real sim as far as the flight model of the KA-50 is concerned. So, why not true dynamic missions and super smart AI ?

1) You think the system requirements are high now, for the above you would be looking at a really good system to run it.

2) It probably would have added about 2 years and a ton of money to the BS release (remember, sim players make up a micro fraction of gamers, most are archade players)

3) After playing it for a couple of weeks, most would give up lol

I play a lot of F4AF, and I solved the problem with the AI back then, the same as I do in BS now..... I DONT USE AI WINGMAN, HUMAN OR NOTHING

I have had the game about 6 weeks, and have yet to play a single player campaign mission. As far as I am concerned, this game was designed for multiplayer, and that is where it shines. I play with REAL people, and that makes EVERY MISSION a totally different ballgame.

Another point, you want complex missions, no problem, make them. Our group creates all our own maps, and with enough triggers you can make it very close to dynamic. Its like everything else in life, you get out of it what you put into it.

Well, thats my 2 cents, and before anyone goes ape on me, remember this. If you want all the hard stuff done for you, then why play a complex sim? pick a game that has everything done for you, like Farcry or Doom.

RedTiger · February 25, 2009

Reading threads like this make me wonder -- am I just too easily amused? My wife would say yes. :D

I haven't even touched the campaigns and rarely ever do. Most of my flight sim time is usually based around flying missions with certain objectives or goals that I want to focus on. I find that I don't need much of a virtual world outside of the pit.

I guess if I could use an example, most of the time I'm just flying stuff to keep up currency. I'm not participating in Red Flag or Mary every single night. I've tried that, but I lose interest fairly quickly. The demands upon my attention are too high. A while back I finally clicked with the campaign in Red Viper and flew several missions nightly for a time, but as soon as something else grabbed my interest, I stopped and I haven't been back. I put aside flight sims for a good while in the past few months, actually. When I picked them up again I didn't have the patience to pick up my Red Viper campaign.

I don't know if this on topic or not, but to get it back on topic, I'll also say that I would also like better AI. I'm talking about BS, I mean I always want better AI since I play off-line. :D

Frederf · February 25, 2009

You don't need hundreds of triggers. Yes, it's a chore - but it's doable. Someone did it for LB2 so you didn't have to.

You might find a mission builder for DCS who'll do the same for you.
Have you actually tried?

I didn't realize that it was my job as the customer to overlook the many missions made by the 1st party and scour the Internet to get something reasonable.

WynnTTr · February 25, 2009

I didn't realize that it was my job as the customer to overlook the many missions made by the 1st party and scour the Internet to get something reasonable.

Not this again Frederf.

You bought the product as is, if you as a customer didn't like it, then don't buy it - either that or personally hire ED to make the game to your specifications.

uhoh7 · February 25, 2009

With a few addtional behaviors, AI would really benifit. E.G. transport helis run away from any threat--logical, but sometimes difficult when making missions. An additional task which would have them attempt to land under fire--ignore threats would be nice.

It would be nice to able to tell wingmen to fire from their present position when in range.

It would also be nice if amphibous vehicles would cross rivers.

Also when a group leader gets confused and stops, so does the whole column e.g. he gets stuck on a side hill. It would be great if the rest of the column would press forward.

Regarding heli AI, they are very fast to obtain targets--especially you. A few more options in this regard would be nice.

Since the sim is all about engaging ground targets and also other helis, AI refinement would be apreciated by everyone.

However that said, current AI is certainly adequate, and not a reason not to buy the sim.

Re GGtharos: he has a point. the ME can produce superb missions, which are anything but "canned". Lack of 3D in the ME means it takes more time to get things placed where you want them, and maybe we could just a few more trigger opttions, espcially a few activated by comms--but there are tons of tools there.

The bottom line is getting more users to turn out missions. I do think ED could do some more outreach in this regard. One issue is the "fear of spaming ethos," which is keeping feedback a bit in check. It takes quite a few hours to turn out a fun mission and there are always things which are missed and need to be updated. Comments for mission designers should be encouraged.

I imagine there are missions being created in the russian community, it would be great to get these translated and available for everyone.

There is really no reason why a healthy community could not be putting out a new SP mission every day--- and this would really boost interest.

Would love to see one from you GG!

MBot · February 25, 2009

Here are my 2 cents:

If AI is of utmost importance to you, Black Shark will most likely not impress you, in some cases even disappoint. It has some impressive features regarding the detection logic of AI air-units, but appart from seeing them at your wingman those will hardly be noticed while playing. The general AI of wingmen is ok but not really spectacular. They are unable to fly close to the ground (about 100m or so is the minimum) or to utilize cover. Sometimes they will show a good display of self-preservation, engaging stand off and refusing to put themselfe in harms way, sometimes they will charge into their doom. Compared to Lock On they will now also call out missiles launches not directed at themselfs, but without more specific information. Generaly the input you get from them about the battlefield is very low (except recon mode wich is a great idea). The wingman AI is very far from the equal tactical partner approach that is so important to NATO air ops (A-10 *cough*).

As attack helicopters should not interract much with aircraft, ground units are the enemys you will interact 99% of the time. Unfortunatly the ground AI is extremly limited. Basicaly they can only follow a set path without pause, drive around obstacles on their path and shoot. Their pre-determined rollcoaster ride can be initiated (but not stopped) by triggers. Otherwise they will not react to actions taken by the player. With the use of the trigger system and the unit activation as consequence, some advanced features can be crudly simulated. But because this will take many trigger lines in the editor, such behaviour is generaly limited to only a few key units. In the my missions you find on this forum I tried to test the limits of the current trigger system.

Like some others I have expierienced that after you master to operate the Ka-50, the game shows some lack of gameplay. Combat and battlefield realism is in no way on the same level as the simulation of the helo. For me the primary motivation to play combat flightsims is to take real world tactics and apply them to the simulation. To be succesfull by using advanced tactics is the climax in sims for me. I just love it when a plan comes together :) Unfortunately in that regard Black Shark is a bit of a shallow experience at the moment. There is not much point to use surprise, confusion or suppression tactics. Trying to remain undetected before attack gives no advantage (and is almost impossible anyway because of the absolute detection logic of ground AI). The only really thing you have to worry about is to keep the correct distance to threats. If you master managing distances (which is not that difficult) you can turn most missions into a shooting range.

Now I said a lot of not so pleasing things about Black Shark as a combat flight simulator. What you have to be aware though is that Black Shark is a spectacular flight simulator. The simulation of the Ka-50 is incredible and is well worth to invest money and time. As such I would highly recommend to buy this sim. Also DCS is in constant development and I am confident that it will evolve into a great 'combat' flight simulation with time. On some of my criticised points they might already work on right now.

Edited February 25, 2009 by MBot

Feuerfalke · February 25, 2009

Very nicely summed up, MBot!

@ GGTharos

I understand your point, GG, but you also got to understand ours:

Nobody claims "Hey, implement randomizer and AI, because it is easy.".

We say "Hey, please implement it, because these are most important features!"

That's an important difference! It doesn't mean that we don't appreciate the product or the tools you provided (or will provide, hopefully). But we point out the weak points that still exist, like it or not.

It's interesting MBot came to the same conclusion I was bashed for on this forums several weeks ago: Right now, we have a flightim, not a combat-sim. That doesn't make the simulation bad, but it takes a lot of motivation out of it. And it can be fixed and hopefully will be, at least with the next module. I don't want to fly CAS for towed target drones. :music_whistling:

CE_Mikemonster · February 25, 2009

Well said MBot, I wish I could have summed it up exactly as you just did.

@ Fuerfalke

Right now, we have a flightim, not a combat-sim. That doesn't make the simulation bad, but it takes a lot of motivation out of it. And it can be fixed and hopefully will be, at least with the next module.

This is all coming back to the same old argument that basically ends up being 'ED produced BS to demonstrate an ability to the military to produce rotor-wing simulations.'

I don't think it's that simple at all, and I very much respect ED for releasing BS.

I enjoy the sim, but it's obvious that the main-effort was to produce the Ka-50, not the 'world' it lives in (and rightly so).

I'm not saying a huge amount of effort didn't go into the AI, i'm sure it did, but nobody could deny that the priority was a fully working virtual Ka-50.

With this in mind, community gripes regarding the missions should have been anticipated by ED, and either a solution given or a straightforward explanation given. I don't think anybody here would have not bought the sim if ED had said 'priority has been given to simulating the Ka-50, AI is still very much WIP.' As a matter of fact I think a lot more people would be 'on side' about it and willing to help.

Tharos, you're not wrong in what you're saying at all, but it could be argued that you're having the same issue again and again (defending this corner) because nobody came clean and just said "we're working on it'."

All i'm doing here is trying to get to the point, it would have happened anyway over the next few pages. I fully respect all involved.

Edited February 25, 2009 by CE_Mikemonster

DiabloSP · February 25, 2009

Although hard, very hard indeed, you can create a realistic combat environment today. Mostly adequate for low intensity conflicts, like the first campaing, but you can do it with a lot of work.

What is being discussed here though, is the fact that MOST of that hard work can be taken off the hands of the mission creator, just by implementing some fairly simple AI routines (fairly simple in most recent games, maybe they are incredibly hard to create in the current BS engine...)

Acedy · February 25, 2009

With this in mind, community gripes regarding the missions should have been anticipated by ED, and either a solution given or a straightforward explanation given. I don't think anybody here would have not bought the sim if ED had said 'priority has been given to simulating the Ka-50, AI is still very much WIP.' As a matter of fact I think a lot more people would be 'on side' about it and willing to help.
Tharos, you're not wrong in what you're saying at all, but it could be argued that you're having the same issue again and again (defending this corner) because nobody came clean and just said "we're working on it'."

Well, please use the search function next time: ED already said multiple times that:

a) aircraft AI has been heavily reworked, as described in the manual and various posts

b) ground unit AI is under development and will be improved. See for example:

http://forums.eagle.ru/showthread.php?p=595701#post595701

http://forums.eagle.ru/showthread.php?p=559334#post559334 (posts 49 and 50)

http://forums.eagle.ru/showpost.php?p=594878&postcount=3

connos · February 25, 2009

I don't care to much about ground units, of course i want their behavior to be updated but i can stand it. What i can't stand is the wingman(you need his missiles to finish some missions) behavior that its not attacking from formation and he has to enter enemy lines and then attack.

Of course you can do it with the datalink (not always works and it's time cosuming) but if you give command engage my target, to a target that i can see and engage why he has to rush towards the target and then engage?

Edited February 25, 2009 by connos

CE_Mikemonster · February 25, 2009

I know that ED are working on the AI and it's going to have a complete overhaul in the future, but saying 'were working on it' isn't the same as saying that it needs work to bring it up to par with the simulation aspect of the game.

Edit: My previous post was a confusing one, some of the context was bad (sorry about that, and for any unintentional annoyance).

What I was trying to say was pretty much what I said above. ED have stated numerous times that they're working on the AI, as with most aspects of the game.

Edited February 25, 2009 by CE_Mikemonster

EvilBivol-1 · February 25, 2009

We've now gotten into the resources discussion, debating ED's priorities in developing Black Shark. I will expand on this a bit...

First of all, developing the Ka-50 model and developing the AI are in general independent tasks. Most of the work going into the aircraft model involves the dynamics programmers that do not otherwise specialize in AI development anyway. If there was no work for them to do on the Ka-50, they would most likely be developing models of other flyables, not AI.

Furthermore, in developing Black Shark, it's important to keep in mind what I would call the basic design concept. Black Shark is hopefully a first release in what ED hopes will be a line of products in which their technology will be continually developed. Much of the work done in Black Shark was with an eye toward the future. Also, remember that ED is trying to develop for the needs of their military contracts as well. Military simulations demand scripting ability over dynamic environments and are certainly not interested in random behavior. Military sims, at least ones ED is contracted for, also benefit from deep modeling of individual units in the virtual world rather than large-scale virtual combat not directly connected to the pilot.

Keeping the design concept in mind, ED put their AI development resources essentially into the following needs for Black Shark: 1) new Mission Editor that is flexible both for the user and for the team to continually develop; 2) re-coding the AI definitions to include individual weapons modeling per units and moving that code to more easily accessible LUA format; 3) re-designing the helicopter AI for stand-off tactics and target prioritization (doesn't mean it's perfect); 4) designing sensor-dependent target detection logic for AI aircraft.

This is certainly not the end of it. If more programmers and/or more time was available, I'm sure other AI features would have been developed as well. As we speak, ED is probably working on new ground AI logic, at least basic self-defense measures when under attack. In the longer-term, of course ED hopes to develop both the ME and AI to a much more powerful and flexible system.

Earlier in this thread, I was just curious to hear what other flight sims have done with AI, especially ground AI compared with Black Shark. IMO, every sim has had to make significant compromises somewhere, but I'm afraid we tend to forget those and focus on the things that worked well. Perhaps small maps? Static unit nodes, player-centered behavior (not independent AI), limiting world "bubbles", etc.?

Edited February 25, 2009 by EvilBivol-1

connos · February 25, 2009

You have good points in you post but i would like a much improve wingman behavior and some simple things like attack from formation.

VMFA117_Poko · February 25, 2009

You have good points in you post but i would like a much improve wingman behavior and some simple things like attack from formation.

If it's the most important thing for you to work on try to fly online with community buddies.

connos · February 25, 2009

I cant i don't have the time for multiplayer. I play in small sessions and i pause frequently some times for an hour just to finish a mission. Multiplayer is not an option for me. Just to be fair i like that your wingman can report airborne threats to you.

GGTharos · February 25, 2009

Very nicely summed up, MBot!

@ GGTharos

I understand your point, GG, but you also got to understand ours:

Nobody claims "Hey, implement randomizer and AI, because it is easy.".
We say "Hey, please implement it, because these are most important features!"

That's an important difference! It doesn't mean that we don't appreciate the product or the tools you provided (or will provide, hopefully). But we point out the weak points that still exist, like it or not.

I understand more than you think, heh :P

I have no idea what ED's plans for more automation are. It probably won't be something that will evolve quickly, but on the other hand, also don't make the mistake of thinking that ED doesn't listen. It's just that there's priorities to set.

It's interesting MBot came to the same conclusion I was bashed for on this forums several weeks ago: Right now, we have a flightim, not a combat-sim. That doesn't make the simulation bad, but it takes a lot of motivation out of it. And it can be fixed and hopefully will be, at least with the next module. I don't want to fly CAS for towed target drones. :music_whistling:

I think DCS is putting the 'Flight' into 'Combat Flight Sim'. You can get the combat part alright, and do keep in mind that things aren't standing still. ;)

BT23 · February 25, 2009

Is the DCS AI better than Flaming Cliffs ?

amalahama · February 25, 2009

Only a reflexion.

Maybe ED is not interested on develop a complex dynamic campaign in DCS.

But what about to provide enough tools to let the community to develop a D.C?

For example, output state files after a mission would be great, to know where units are and what's their state (damage, destroyed, healthy...). Some kind of algorythm to let lots of AI items to combat with few fps loss (something like 2D Falcon's bubble world) should be desirable. D.C logic and automatic mission planner could be developed by the community.

Regards!!

Edited February 25, 2009 by amalahama

MBot · February 25, 2009

Earlier in this thread, I was just curious to hear what other flight sims have done with AI, especially ground AI compared with Black Shark. IMO, every sim has had to make significant compromises somewhere, but I'm afraid we tend to forget those and focus on the things that worked well. Perhaps small maps? Static unit nodes, player-centered behavior (not independent AI), limiting world "bubbles", etc.?

I think this aspect deserves some thought. Actually I can't name a sim that had great AI overall. Some have left a good impression, but I have to say that I had not that in-depth look at things back then that I have now. Gunship! for example left a good impression of a capable ground AI, but I can't provide any details or data. Perhaps it was only overshadowed by a generally good representation of mechanized warfare.

Regarding sophistication of AI, Black Shark is therefore most likely not much different than most others sims. The thing is that while AI did not significantly advanced in the last 10 years, I personally did. I master sims much quicker and developed the habit to analyse them in detail. As such I now notice shortcomings that I would have overlooked 10 years ago. And the thing I notice most often is that AI, in my opinion the key to good gameplay, today lags behind a lot in the total game package (be it in sims, tactical-shooters or most other games). AI was not really better back then, but it fit better into the overall development-level of games and my expectations. Today, looking at the progress graphics, physics, flight models and system depth have made, I can't help but to notice that AI is underdeveloped.

hawk66 · February 25, 2009

Oh :),

I've expected anything but not that amount of feedback. Thanks guys!

Just on remark to the AI of Falcon4AF(latest patch): I think it has a pretty good, dynamic AI...by no means it is perfect, but it gives you sometimes the feeling that you are really in a virtual fighter/environment.

It's true...you can't make anybody happy. I see the point that the focus of this sim is the helicopter itself. But I just do not like 'sterile' simulations.

I've skimmed the manual for the mission editor and it looks that it has a lot of functions.

But my experience is: You can have such tools but it will by no means a substitute for a dynamic campaign/AI like Falcon4 or even a very old sim like RedStormRising has.

I do not know if anybody knows the naval sim 'Dangerous Waters'. It has a very sophisticated editor, a couple of random functions (probably even more than Black Shark has). But until now I haven't seen a very good, dynamic campaign which is replayable. I've tried it by myself: You try to create a complex, dynamic scenario(that means you use a lot random functions/triggers etc.) but:

there is no automatic testing like unit testing in programming languages. You change sth and you often have to test the whole stuff again
you basically can only create simple rules in the form IF....ELSE...ENDIF but sometimes you need to use more advanced algorithms like fuzzy logic or even simple loops.
it's a huge amount of work
but the most important issue: you just simply loose the overview. You have hundreds of triggers/functions available and they are all connected because you try to build a dynamic mission/campaign. But you, as a human, don't have a chance to handle this complexity. Try to create 100 rules to filter emails in your favorite email program. Are you already lost? But that's just a piece of cake in comparison to create a dynamic scneario.

There's another sim: Steel Beasts pro (tank sim). In this sim the approach with the mission editor works because the goal is to create (reproducable (but a little randomized) training missions for the military and no campaigns.

I'll wait until the patch to make my final decision but nevertheless I'll really appreciate that there is now a producer which creates high fidelity simulations. So this post is no bashing but just to express my personal opinion!.

Edited February 25, 2009 by hawk66

AlphaInfinity · February 25, 2009

Guys, visit us on the ka50.com dedicated server. We are running some missions by TX-Kingsnake. The MP missions are excellent and I have a feeling that this is only a preview of what the potential could be in the future. Black Shark is pretty new and a lot of people have not had the chance to really dig into the mission editor and yet we already have some EXCELLENT MP missions out there.

EtherealN · February 25, 2009

For example, output state files after a mission would be great, to know where units are and what's their state (damage, destroyed, healthy...). Some kind of algorythm to let lots of AI items to combat with few fps loss (something like 2D Falcon's bubble world) should be desirable. D.C logic and automatic mission planner could be developed by the community.

I think this would be relatively simple, but I don't know the codebase so couple shovels of salt on that, but a good intermediate step would be to have the ability to make triggers set variables that are carried on and fed into the next campaign mission - which would then obviously have to be made to accept those variables and the person making the mission would have to do that manually.

This won't make anything "dynamic" but it would facilitate the "smoke and mirrors" when making campaigns. And from years of playing IL-2 in dynamic campaign mode I have to say I'd much rather have well made smoke and mirrors than any dynamic campaign I have ever seen. (Even though IL-2 did a passable job. Passable.) I would assume that it would also be a relatively simple thing to implement compared to a complete revision of AI and all those other things that have been proposed here.

Step after that would of course be to get to write in logic - effectively more advanced scripting - into the missions. But I suspect that that's where it starts getting really complex. :P

Oh, and being able to create custom radio commands would be awesome, like in the old Flashpoint editor. That way we could actually "tell" the ground group we are escorting to stop or go or run like sissies etcetera.

H-street · February 25, 2009

after reading this it now makes much more sense why ED is not interesting in making a dynamic campaign...

but it seems the complaints about hte AI and a Dynamic Campaign are one in the same. Making the world feel alive and allowing us to immerse ourselves in it.

In that sense this has been hashed out many times in the dynamic campaign discussions. This idea of building a "world" instead of building just a simulator. There is a reason Faclon 4.0 is still so popular today amongst combat simmers, and that is its basic fundamental design principle of ..

not just to simulate the aircraft but the entire fighter pilot experience.

for me Blackshark is fun for flipping switches and flying around, shooting at a few things.. many times i find i enjoy just a joyride.. but not like in Falcon where everytime I go up there is a sense of a living world (even with its flaws)..

with that said, BS is what it is. and from what i have seen indicate that DCS this way for a while, even with the A10 module coming out i haven't seen much interest from them in making that "living" world

My gut instinct is that since DCS wasn't built from the ground up with this living world in mind will make it much more difficult to achieve.

Sign In

AI (decision to buy this sim)

Recommended Posts

rapid

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

FADM Stern GNSF

RedTiger

Frederf

WynnTTr

uhoh7

MBot

Feuerfalke

CE_Mikemonster

DiabloSP

Acedy

connos

CE_Mikemonster

EvilBivol-1

connos

VMFA117_Poko

connos

GGTharos

BT23

amalahama

MBot

hawk66

AlphaInfinity

EtherealN

H-street

Recently Browsing 0 members